And with Xinerama disabled, running glxinfo on the second head (r128) makes the server segfault! That screen (:0.1) wasn't using DRI. I think the root cause was that the r128 kernel module wasn't loaded. Maybe there's another bug here somewhere, since X should cause the dri module to load like it does for i915 for the first head.
The server's i810_drv.so is v1.7.3, that I compiled myself, and the client libGL and libdrm are also the latest git versions that I compiled locally. The only files outside of /usr/local that are locally compiled are i810_drv.so, and /usr/X11R6/lib/modules/dri. No, that isn't where ubuntu puts r128_dri.so, but that's where mesa 6.5.2 in /usr/local/lib looks for them... Err, /usr/lib/dri/* is from libgl1-mesa-dri, which I built locally from Ubuntu source, because I needed to compile with -fno-strict-aliasing for the i965 driver to work. Anyway, I'll attach my Xorg log file in the hope that it's of some use even though it's not with the Ubuntu versions of everything. It's a big pain to use Edgy's i810_drv.so, because it screws up the vid mode. I'll send this now before I do anything that risks crashing X and losing what I've typed in firefox, since I'm not running it in Xvnc or anything! ** Attachment added: "Xorg log from glxinfo crash" http://librarian.launchpad.net/5392246/Xorg.0.log.glxinfo-r128-second-head.crash -- libGLcore crash (probably screensaver-induced) https://launchpad.net/bugs/75592 -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
